我编写了一个mapreduce程序（mapper.py和reducer.py）来处理hadoop中的pagerank问题。
我想迭代mapreduce大约10次。如何将第一轮mapreduce的输出转换为第二轮mapreduce的输入？

1                    2                           10
[mapper->reducer] -> [mapper->reducer] -> ... -> [mapper->reducer] -> final result